Google is opening up Spanner—a massive database used internally at the search giant—to outside users. The move had been rumored for some time, and could be an important step forward in Google's plans to challenge Amazon Web Services in the public cloud market.
Starting Tuesday, customers can try out a test or beta version of Spanner.
"We think this is the first global scale distributed SQL database," Google Cloud vice president Brian Stevens tells Fortune.
The key point: Spanner can be spread across not only multiple machines but multiple data centers, and yet somehow maintain that the data across all those locations is synchronized—or always consistent.
That is a very tough problem that until now appeared unsolvable by a relational database. That difficulty led to the boom in NoSQL databases that promised "eventual" data consistency. Examples of NoSQL databases include MongoDB (mongodb), Cassandra, Hbase. Google (googl) itself fields a NoSQL database called Bigtable.
SQL stands for the Structured Query Language, commonly used to interact with what are called relational databases. NoSQL, by contrast, stands for Not Only SQL or not-SQL, meaning that NoSQL databases are set up and queried differently.
If your company is a huge retailer with thousands of stores worldwide, it would be helpful to know exactly what the product inventory situation is at any one time.
"With Spanner, you click on one database, and you'll know what products are selling and where," says Google Cloud product manager Deepti Srivastava.
Spanner pricing is based on number of compute nodes, amount of storage, and network bandwidth used.
Get Data Sheet, Fortune’s daily newsletter on the business of technology.
Five years ago, Google published some details about Spanner, which tech news site Gigaom reported on at the time. Google is offering customers the same infrastructure and service it is running internally, Srivastava notes.
That is no small point. Google has huge tech capabilities gleaned from running its own Internet search and ad business. But most of the cloud technology it's rolled out to customers as part of the Google Cloud Platform is not exactly the same as what runs internally. The goal, however, remains for those two sets of technology to find each other.
For more on Google Cloud watch:
JDA Software, a company that sells supply chain management software, is among a few customers who have had early access to Spanner. John Sarvari, group vice president of technology at JDA, says the ability to guarantee data consistency across locations is huge for the vendor.
While JDA will continue to use NoSQL databases for some things, there are applications where this consistency is paramount. "With today's NoSQL technologies, there are many things we have to do in engineering to overcome the limitations of that eventual consistency model. That requirement goes away with Spanner," Sarvari says.
If Google can bring out its own technology in a format useable by companies that don't necessarily work at Google's scale, it could help its case against AWS (amzn) and the number two contender, Microsoft Azure.
Date: (February 15, 2017 8:15 a.m.) This story was updated to add links to the Google blog and Spanner pricing.